honey::MinimizeN< Real, Dim > Class Template Reference

Find a local minimum of an n-dimensional function using "Powell's conjugate gradient descent method". More...

#include <MinimizeN.h>

Public Types
typedef Vec< Dim, Real >	Vec
typedef function< Real(Vec)>	Func

Public Member Functions
	MinimizeN (Real tol=Real_::zeroTol, int iterMax=30, int levelMax=30, int bracketMax=30)
tuple< Vec, Real >	calc (const Func &func, const Vec &min, const Vec &max, const Vec &init)
	Find the minimum of a function within bounds [min,max] using init as an initial guess. More...

Static Public Attributes
static const int	dim = Dim

Detailed Description

template<class Real, int Dim>
class honey::MinimizeN< Real, Dim >

Find a local minimum of an n-dimensional function using "Powell's conjugate gradient descent method".

Member Typedef Documentation

template<class Real , int Dim>

typedef function<Real (Vec)> honey::MinimizeN< Real, Dim >::Func

template<class Real , int Dim>

typedef Vec<Dim, Real> honey::MinimizeN< Real, Dim >::Vec

Constructor & Destructor Documentation

template<class Real , int Dim>

honey::MinimizeN< Real, Dim >::MinimizeN ( Real  tol = Real_::zeroTol, int  iterMax = 30, int  levelMax = 30, int  bracketMax = 30  )

inline

Parameters

tol	find minimum to within tolerance
iterMax	max number of iterations for gradient descent method
levelMax	see Minimize
bracketMax	see Minimize

Member Function Documentation

template<class Real , int Dim>

auto honey::MinimizeN< Real, Dim >::calc	(	const Func &	func,
		const Vec &	min,
		const Vec &	max,
		const Vec &	init
	)

Find the minimum of a function within bounds [min,max] using init as an initial guess.

The algorithm starts at point init and searches for a minimum along each coordinate axis using the Minimize class. A "conjugate" vector is then created by adding the deltas along each axis. A new starting point is then found by taking the minimum along the conjugate vector. The most successful axis is then replaced with the conjugate to minimize linear dependency. The direction set is cycled and the algorithm repeats to iterMax.

Parameters

func
min	minimum lower bound
max	minimum upper bound
init	initial guess of minimum

Return values

argMin	the function arg that results in the minimum
valMin	the minimum

Member Data Documentation

template<class Real , int Dim>

const int honey::MinimizeN< Real, Dim >::dim = Dim

static

---

The documentation for this class was generated from the following files:

• src/common/Honey/Math/NumAnalysis/MinimizeN.h
• src/common/Honey/Math/NumAnalysis/MinimizeN.cpp